一种基于改进LCS的多移动机器人学习算法

移动机器人采用学习分类器系统进行学习的主要问题在于学习过程较长。为了解决该问题，本文提出一种改进的学习分类器系统。该系统引入了一个规则构造器和合并、广播两个操作。规则构造器在学习分类器系统进行新的尝试时产生新的规则，合并操作对已有规则进行归纳，广播操作则使所有的机器人可以共享各自当前的最优规则。仿真结果表明这些措施有效地提高了学习分类器系统的收敛速度。

A Learning Algorithm Applied to Multiple Mobile Robots System Based on Improved LCS

The main problem of Learning Classifier System (LCS) applied to mobile robots is the low speed of convergence. An improved LCS is proposed to solve this problem. A Rule Constructor and two new operations: Merge and Broadcast are introduced. The Rule Constructor produces new classifiers when LCS takes a try; the Merge operation sums up the better classifiers and the Broadcast operation enables the robots to share their best classifiers. The simulation results demonstrate that these measures obviously accelerate the procedure of learning.